Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The State of Robot Motion Generation

Kostas E. Bekris, Joe Doerr, Patrick Meng, Sumanth Tangirala

TL;DR

The paper surveys a broad landscape of robot motion generation methods, contrasting explicit-model planning with data-driven implicit models. It highlights the maturity and reliability of explicit-model approaches while recognizing the growing capabilities and data requirements of learning-based methods. A central message is that integrative architectures combining planning, perception, and verification, together with safe deployment practices, are essential for real-world robotics. By identifying gaps in interfaces, benchmarks, and cross-community collaboration, the work guides future research toward robust, scalable, and trustworthy motion generation systems.

Abstract

This paper reviews the large spectrum of methods for generating robot motion proposed over the 50 years of robotics research culminating in recent developments. It crosses the boundaries of methodologies, typically not surveyed together, from those that operate over explicit models to those that learn implicit ones. The paper discusses the current state-of-the-art as well as properties of varying methodologies, highlighting opportunities for integration.

The State of Robot Motion Generation

TL;DR

The paper surveys a broad landscape of robot motion generation methods, contrasting explicit-model planning with data-driven implicit models. It highlights the maturity and reliability of explicit-model approaches while recognizing the growing capabilities and data requirements of learning-based methods. A central message is that integrative architectures combining planning, perception, and verification, together with safe deployment practices, are essential for real-world robotics. By identifying gaps in interfaces, benchmarks, and cross-community collaboration, the work guides future research toward robust, scalable, and trustworthy motion generation systems.

Abstract

This paper reviews the large spectrum of methods for generating robot motion proposed over the 50 years of robotics research culminating in recent developments. It crosses the boundaries of methodologies, typically not surveyed together, from those that operate over explicit models to those that learn implicit ones. The paper discusses the current state-of-the-art as well as properties of varying methodologies, highlighting opportunities for integration.

Paper Structure

This paper contains 12 sections, 2 figures.

Table of Contents

  1. Introduction
  2. Motion Generation given Explicit Models
  3. Motion Planning
  4. Task and Motion Planning (TAMP)
  5. Belief Space Planning
  6. Control and Feedback-based Planning
  7. Data-driven Motion Generation with Implicit Models
  8. Learning from Demonstrations
  9. Reinforcement Learning (RL)
  10. Cross-task Learning
  11. Large Models
  12. Perspectives on Robot Motion Generation Research

Figures (2)

  • Figure 1: Robot motion generation methods that operate over an explicit model.
  • Figure 2: Robot motion generation methods that operate over an implicit model.